3 ms·
Makes sense, and each model has a max context length, so they could charge per token assuming full context by model if they wanted to assume worst case.
by daxfohl 1y ago
Makes sense, and each model has a max context length, so they could charge per token assuming full context by model if they wanted to assume worst case.